96 ZOROASTERIDAE. 



While Mammaster remains monotypic, the following species are to be referred 

 to Cnemidaster: — wyvillei, squameus, zea, nudus. 



At the same time that he proposed Mammaster, Perrier also seems to pro- 

 pose a new genus, Prognaster, with Z. longicauda as the type. He had however, 

 three years earUer (1891. Mem. Soc. Zool. France, 4, p. 259) given the same 

 name and diagnosis to a genus with a new species, grimaldii, as the type. This 

 earher pubUcation of the name is only referred to casually and ambiguously in 

 the later paper, the user of the Travailleur and Talisman Report being thus 

 completely and quite needlessly misled. The discovery of the many long- 

 armed Zoroasters of the Indian Ocean has made the proposed genus as repre- 

 sented by longicaudus quite untenable and I am not able to find any satisfactory 

 basis for separating that species frona the more typical species of Zoroaster. 

 But if Perrier is correct in his statement regarding grimaldii that the adambula- 

 cral plates are aU ahke, the genus Prognaster may well be retained for so unique 

 a zoroasterid. 



In 1905, Fisher (Bull. U. S. Bur. Fish., 24, p. 316) proposed a subgenus 

 Myxoderma for two new Zoroasters, sacculatus and evermanni, designating 

 sacculatus as the type. Tliis group, to which platy acanthus H. L. Clark also 

 belongs, is well-characterized and Fisher has recently (1919. Ann. Mag. Nat. 

 Hist., ser. 9, 3, p. 387) given it full generic rank. But evermanni proves to be 

 a true Zoroaster. Another genus described by Fisher (1916. Proc. Biol. Soc. 

 Washington, 29, p. 31) is Bythiolophus, based on material collected by the 

 Albatross in Buton Strait, Celebes. This seems to be quite distinct from 

 anytliing hitherto known in the family. Finally mention must be made of 

 Sladen's (1889. Challenger Ast. p. 426) genus PhoUdaster, recently rediscovered 

 by the Albatross. 



There are then seven apparently valid genera of Zoroasteridae. 



Key to the genera of Zoroasteridae. 



A. Primary plates and usually additional disk-plates conspicuously large, convex (often swollen), 

 and more or less bare. 

 B. First carinal plate, just distal to primary radial, conspicuously enlarged, larger than the adjoin- 

 ing second marginals which are very much larger than the first ones; adradials wanting; three 



series of actinolaterals with beginnings of a fourth in large specimens Mammaster. 



BB. First carinal scarcely or not at all larger than the first marginals which are evidently larger than 



the second; adradials present; four or five series of actinolateral plates Cnemidaster. 



AA. Disk-plates not conspicuously convex or swollen, or it so, closely covered with spinules. 



B.^ Adambulacral plates all alike Prognaster. 



BB.' Every other adambulacral plate with a conspicuous ridge projecting into furrow and carrying 

 long spines. 



